A Feature Extraction Algorithm for Welding Seams with Robustness to Disturbance
An Image feature extraction algorithm with robustness to arc light and reflex light for welding seams is proposed in this paper. The feature extraction algorithm consists of image preprocessing and extracting feature points. Computing target area and enhancing image are finished during preprocessing process. Target area is determined by computing the sum of pixel errors in a moveable data window. Image enhancement is based on adaptive thresholds which are gained by computing the frequency of gray level in target area. A data set separation algorithm based on clustering method is designed to extract features of welding seams in target area. The algorithm puts pixels into different line data sets, and then a line fitting method with weighted values is designed to improve the fitting precision. Weighted values are decided according to the level of validity of pixel coordinate and the distance between pixel to be checked and the intersection point of two adjacent data sets. Experiments verify the validity of algorithms proposed in this paper.
Yan Wang Linkun Wang De Xu
Beihang University Beijing, China 100083 Instrumentation Technology and Economy Institute Beijing, China 100083 Institute of Automation, Chinese Academy of Sciences Beijing, China 100083
国际会议
南宁
英文
2007-07-20（万方平台首次上网日期，不代表论文的发表时间）